Skip to contentSkip to conversation

ieee802154: rework receive handling

PATCH In review Alexander Aring 2 participants 18 messages Oct 27, 2014

Subsystems (5)
lore
Share & tools
lore

Alexander Aring

author maintainer
#0 lore raw

v2 cover ieee802154: rework receive handling

This patch serie improves and cleanups the receive handling. Instead workqueue we use a tasklet now for parsing frames. The first patch is a fix for using put_unaligned_le16 to copy checksum in skb data which was reported by Marcel Holtmann.

Other patches are cleanups for preparing further patches to introduce a frame parsing like mac80211. We don't change the printout function to netdev_foo now because the most code will be replaced by the new frame parsing implementation.

changes since v2:
- correct ieee802154 instead ieee80211 in patch 04/16 ("mac802154: rx:
document ieee802154_rx() context requirement")
- reword complete commit message of patch 04/16 ("mac802154: rx: document ieee802154_rx() context requirement")
- fix && instead || in patch 13/16 ("ieee802154: add valid psdu length helper")
- make patch 13/16 ("ieee802154: add valid psdu length helper") more human readable to check start at first end then end interval. Instead end <-> start checking.

Alexander Aring (16):
mac802154: tx: use put_unaligned_le16 for copy crc
ieee802154: drivers: use dev_alloc_skb
mac802154: rx: use tasklet instead workqueue
mac802154: rx: document ieee802154_rx() context requirement
mac802154: rx: move receive handling into rx.c
mac802154: tx: remove monitor receive while xmit
mac802154: rx: rename remove mac802154_subif_rx
mac802154: rx: move skb->protocol setting
mac802154: rx: add CHECKSUM_UNNECESSARY
mac802154: rx: add monitor pkt_type information
mac802154: rx: move skb_reset_mac_header
mac802154: rx: move rcu locking
ieee802154: add valid psdu length helper
at86rf230: use ieee802154_is_valid_psdu_len helper
at86rf230: improve receive handling
mac802154: rx: change naming convention

 drivers/net/ieee802154/at86rf230.c |  32 ++---
 drivers/net/ieee802154/cc2520.c    |   2 +-
 drivers/net/ieee802154/mrf24j40.c  |   2 +-
 include/linux/ieee802154.h         |  11 ++
 include/net/mac802154.h            |   1 +
 net/mac802154/ieee802154_i.h       |   9 +-
 net/mac802154/iface.c              | 186 -------------------------
 net/mac802154/main.c               |  30 ++++
 net/mac802154/monitor.c            |  27 ----
 net/mac802154/rx.c                 | 279 +++++++++++++++++++++++++++++++------
 net/mac802154/tx.c                 |   7 +-
 11 files changed, 296 insertions(+), 290 deletions(-)
Signature · 1 line
2.1.2
author maintainer
#1 lore raw

PATCH v2 1/16 mac802154: tx: use put_unaligned_le16 for copy crc

This patch replaces the memcpy with a put_unaligned_le16. The placement of crc inside of PSDU can also be unaligned. With memcpy this can fail on some architectures.

---
 net/mac802154/tx.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
net/mac802154/tx.c +3 -2
index fe2e17e..31e51e4 100644
@@ -20,6 +20,7 @@
 #include <linux/netdevice.h>
 #include <linux/if_arp.h>
 #include <linux/crc-ccitt.h>
+#include <asm/unaligned.h>
 
 #include <net/rtnetlink.h>
 #include <net/ieee802154_netdev.h>
@@ -84,9 +85,9 @@ ieee802154_tx(struct ieee802154_local *local, struct sk_buff *skb)
 	mac802154_monitors_rx(local, skb);
 
 	if (!(local->hw.flags & IEEE802154_HW_OMIT_CKSUM)) {
-		__le16 crc = cpu_to_le16(crc_ccitt(0, skb->data, skb->len));
+		u16 crc = crc_ccitt(0, skb->data, skb->len);
 
-		memcpy(skb_put(skb, 2), &crc, 2);
+		put_unaligned_le16(crc, skb_put(skb, 2));
 	}
 
 	if (skb_cow_head(skb, local->hw.extra_tx_headroom))
Signature · 1 line
2.1.2
author maintainer
#2 lore raw

PATCH v2 2/16 ieee802154: drivers: use dev_alloc_skb

In reply to Alexander Aring

This patch change the allocation of skb inside the ieee802154 driver layer to dev_alloc_skb. This changes also the gfp mask to GFP_ATOMIC which is needed for upcomming change that the receiving is done by a tasklet and not a workqueue anymore.

---
 drivers/net/ieee802154/at86rf230.c | 2 +-
 drivers/net/ieee802154/cc2520.c    | 2 +-
 drivers/net/ieee802154/mrf24j40.c  | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)
drivers/net/ieee802154/at86rf230.c +1 -1
index 0669783..e4fbcaa 100644
@@ -811,7 +811,7 @@ at86rf230_rx(struct at86rf230_local *lp,
 	memcpy(rx_local_buf, data, len);
 	enable_irq(lp->spi->irq);
 
-	skb = alloc_skb(IEEE802154_MTU, GFP_ATOMIC);
+	skb = dev_alloc_skb(IEEE802154_MTU);
 	if (!skb) {
 		dev_vdbg(&lp->spi->dev, "failed to allocate sk_buff\n");
 		return;
drivers/net/ieee802154/cc2520.c +1 -1
index f6f07f4..a31b5b6 100644
@@ -524,7 +524,7 @@ static int cc2520_rx(struct cc2520_private *priv)
 	if (len < 2 || len > IEEE802154_MTU)
 		return -EINVAL;
 
-	skb = alloc_skb(len, GFP_KERNEL);
+	skb = dev_alloc_skb(len);
 	if (!skb)
 		return -ENOMEM;
 
drivers/net/ieee802154/mrf24j40.c +1 -1
index 3d775af..7abb237 100644
@@ -544,7 +544,7 @@ static int mrf24j40_handle_rx(struct mrf24j40 *devrec)
 	val |= 4; /* SET RXDECINV */
 	write_short_reg(devrec, REG_BBREG1, val);
 
-	skb = alloc_skb(len, GFP_KERNEL);
+	skb = dev_alloc_skb(len);
 	if (!skb) {
 		ret = -ENOMEM;
 		goto out;
Signature · 1 line
2.1.2
author maintainer
#3 lore raw

PATCH v2 3/16 mac802154: rx: use tasklet instead workqueue

In reply to Alexander Aring

Tasklets have much less overhead than workqueues. This patch also removes the heap allocation for the worker on receiving path. Like mac80211 we should prefer use a tasklet here instead a workqueue to getting fast out of interrupt context when ieee802154_rx_irqsafe is called by driver. Like wireless inside the tasklet context we should call netif_receive_skb instead netif_rx_ni anymore.

---
 include/net/mac802154.h      |  1 +
 net/mac802154/ieee802154_i.h |  7 +++++++
 net/mac802154/iface.c        |  2 +-
 net/mac802154/main.c         | 30 +++++++++++++++++++++++++++
 net/mac802154/rx.c           | 48 ++++++++------------------------------------
 5 files changed, 47 insertions(+), 41 deletions(-)
include/net/mac802154.h +1 -0
index 942dd53..4c4642e 100644
@@ -202,6 +202,7 @@ void ieee802154_free_hw(struct ieee802154_hw *hw);
 int ieee802154_register_hw(struct ieee802154_hw *hw);
 void ieee802154_unregister_hw(struct ieee802154_hw *hw);
 
+void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb);
 void ieee802154_rx_irqsafe(struct ieee802154_hw *hw, struct sk_buff *skb,
 			   u8 lqi);
 
net/mac802154/ieee802154_i.h +7 -0
index ef29c10..603509a 100644
@@ -55,11 +55,18 @@ struct ieee802154_local {
 	 * read them using any of protection methods.
 	 */
 	bool running;
+
+	struct tasklet_struct tasklet;
+	struct sk_buff_head skb_queue;
 };
 
 #define	MAC802154_DEVICE_STOPPED	0x00
 #define MAC802154_DEVICE_RUN		0x01
 
+enum {
+	IEEE802154_RX_MSG        = 1,
+};
+
 /* Slave interface definition.
  *
  * Slaves represent typical network interfaces available from userspace.
net/mac802154/iface.c +1 -1
index be45dc9..311f60c 100644
@@ -376,7 +376,7 @@ void mac802154_wpan_setup(struct net_device *dev)
 
 static int mac802154_process_data(struct net_device *dev, struct sk_buff *skb)
 {
-	return netif_rx_ni(skb);
+	return netif_receive_skb(skb);
 }
 
 static int
net/mac802154/main.c +30 -0
index 3c0a824..ff0de0f 100644
@@ -222,6 +222,29 @@ static int mac802154_set_frame_retries(struct wpan_phy *phy, s8 retries)
 	return local->ops->set_frame_retries(&local->hw, retries);
 }
 
+static void ieee802154_tasklet_handler(unsigned long data)
+{
+	struct ieee802154_local *local = (struct ieee802154_local *)data;
+	struct sk_buff *skb;
+
+	while ((skb = skb_dequeue(&local->skb_queue))) {
+		switch (skb->pkt_type) {
+		case IEEE802154_RX_MSG:
+			/* Clear skb->pkt_type in order to not confuse kernel
+			 * netstack.
+			 */
+			skb->pkt_type = 0;
+			ieee802154_rx(&local->hw, skb);
+			break;
+		default:
+			WARN(1, "mac802154: Packet is of unknown type %d\n",
+			     skb->pkt_type);
+			kfree_skb(skb);
+			break;
+		}
+	}
+}
+
 struct ieee802154_hw *
 ieee802154_alloc_hw(size_t priv_data_len, struct ieee802154_ops *ops)
 {
@@ -270,6 +293,12 @@ ieee802154_alloc_hw(size_t priv_data_len, struct ieee802154_ops *ops)
 	INIT_LIST_HEAD(&local->interfaces);
 	mutex_init(&local->iflist_mtx);
 
+	tasklet_init(&local->tasklet,
+		     ieee802154_tasklet_handler,
+		     (unsigned long)local);
+
+	skb_queue_head_init(&local->skb_queue);
+
 	return &local->hw;
 }
 EXPORT_SYMBOL(ieee802154_alloc_hw);
@@ -371,6 +400,7 @@ void ieee802154_unregister_hw(struct ieee802154_hw *hw)
 	struct ieee802154_local *local = hw_to_local(hw);
 	struct ieee802154_sub_if_data *sdata, *next;
 
+	tasklet_kill(&local->tasklet);
 	flush_workqueue(local->workqueue);
 	destroy_workqueue(local->workqueue);
 
net/mac802154/rx.c +8 -40

77 lines not on this page · raw message

Signature · 1 line
2.1.2
author maintainer
#4 lore raw

PATCH v2 4/16 mac802154: rx: document ieee802154_rx() context requirement

In reply to Alexander Aring

This patch is similar like d20ef63d32461332958661df73e21c0ca42601b0 ("mac80211: document ieee80211_rx() context requirement"). The netif_receive_skb call requires with softirqs disabled. This patch adds a warning if softirqs are pending while calling ieee802154_rx.

---
 net/mac802154/rx.c | 2 ++
 1 file changed, 2 insertions(+)
net/mac802154/rx.c +2 -0
index 2851a3f..c4df321 100644
@@ -61,6 +61,8 @@ fail:
 
 void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
 {
+	WARN_ON_ONCE(softirq_count() == 0);
+
 	mac802154_subif_rx(hw, skb);
 }
 EXPORT_SYMBOL(ieee802154_rx);
Signature · 1 line
2.1.2
author maintainer
#5 lore raw

PATCH v2 5/16 mac802154: rx: move receive handling into rx.c

In reply to Alexander Aring

This patch removes all relevant receiving functions inclusive frame parsing into rx file. Like mac80211 we should implement the complete receive handling and parsing in this file.

---
 net/mac802154/ieee802154_i.h |   1 -
 net/mac802154/iface.c        | 186 -------------------------------------
 net/mac802154/monitor.c      |  27 ------
 net/mac802154/rx.c           | 214 +++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 214 insertions(+), 214 deletions(-)
net/mac802154/ieee802154_i.h +0 -1
index 603509a..be2f2f6 100644
@@ -131,7 +131,6 @@ void mac802154_monitor_setup(struct net_device *dev);
 netdev_tx_t
 ieee802154_monitor_start_xmit(struct sk_buff *skb, struct net_device *dev);
 
-void mac802154_wpans_rx(struct ieee802154_local *local, struct sk_buff *skb);
 void mac802154_wpan_setup(struct net_device *dev);
 netdev_tx_t
 ieee802154_subif_start_xmit(struct sk_buff *skb, struct net_device *dev);
net/mac802154/iface.c +0 -186

189 lines not on this page · raw message

net/mac802154/monitor.c +0 -27
index 5758322..dfdedc2 100644
@@ -18,9 +18,7 @@
  */
 
 #include <linux/netdevice.h>
-#include <linux/skbuff.h>
 #include <linux/if_arp.h>
-#include <linux/crc-ccitt.h>
 #include <linux/ieee802154.h>
 
 #include <net/mac802154.h>
@@ -30,31 +28,6 @@
 
 #include "ieee802154_i.h"
 
-void mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
-{
-	struct sk_buff *skb2;
-	struct ieee802154_sub_if_data *sdata;
-	u16 crc = crc_ccitt(0, skb->data, skb->len);
-	u8 *data;
-
-	rcu_read_lock();
-	list_for_each_entry_rcu(sdata, &local->interfaces, list) {
-		if (sdata->type != IEEE802154_DEV_MONITOR ||
-		    !netif_running(sdata->dev))
-			continue;
-
-		skb2 = skb_clone(skb, GFP_ATOMIC);
-		skb2->dev = sdata->dev;
-		skb2->pkt_type = PACKET_HOST;
-		data = skb_put(skb2, 2);
-		data[0] = crc & 0xff;
-		data[1] = crc >> 8;
-
-		netif_rx_ni(skb2);
-	}
-	rcu_read_unlock();
-}
-
 static const struct net_device_ops mac802154_monitor_ops = {
 	.ndo_open		= mac802154_slave_open,
 	.ndo_stop		= mac802154_slave_close,
net/mac802154/rx.c +214 -0

223 lines not on this page · raw message

Signature · 1 line
2.1.2
author maintainer
#6 lore raw

PATCH v2 6/16 mac802154: tx: remove monitor receive while xmit

In reply to Alexander Aring

This removes the call of monitor receive funktion when any interface type call xmit. There exist no such use case that a monitor interface should receive the actual sending frame. One use case could be that a wpan interface and monitor interface could be running at the same time on one phy. Then the monitor interface receives the wpan frames also. Furthermore we adding support for promiscous mode setting. With promiscous mode setting we can't run a wpan and monitor interface at the same time.

---
 net/mac802154/ieee802154_i.h | 1 -
 net/mac802154/rx.c           | 3 ++-
 net/mac802154/tx.c           | 2 --
 3 files changed, 2 insertions(+), 4 deletions(-)
net/mac802154/ieee802154_i.h +0 -1
index be2f2f6..ac907d9 100644
@@ -126,7 +126,6 @@ extern struct ieee802154_mlme_ops mac802154_mlme_wpan;
 int mac802154_slave_open(struct net_device *dev);
 int mac802154_slave_close(struct net_device *dev);
 
-void mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb);
 void mac802154_monitor_setup(struct net_device *dev);
 netdev_tx_t
 ieee802154_monitor_start_xmit(struct sk_buff *skb, struct net_device *dev);
net/mac802154/rx.c +2 -1
index d8498c5..04f3d61 100644
@@ -216,7 +216,8 @@ mac802154_wpans_rx(struct ieee802154_local *local, struct sk_buff *skb)
 		kfree_skb(skb);
 }
 
-void mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
+static void
+mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 {
 	struct sk_buff *skb2;
 	struct ieee802154_sub_if_data *sdata;
net/mac802154/tx.c +0 -2
index 31e51e4..e857673 100644
@@ -82,8 +82,6 @@ ieee802154_tx(struct ieee802154_local *local, struct sk_buff *skb)
 	struct net_device *dev = skb->dev;
 	int ret;
 
-	mac802154_monitors_rx(local, skb);
-
 	if (!(local->hw.flags & IEEE802154_HW_OMIT_CKSUM)) {
 		u16 crc = crc_ccitt(0, skb->data, skb->len);
 
Signature · 1 line
2.1.2
author maintainer
#7 lore raw

PATCH v2 7/16 mac802154: rx: rename remove mac802154_subif_rx

In reply to Alexander Aring

This patch removes the mac802154_subif_rx function and do the necessary calls inside of ieee802154_rx function. The ieee802154_rx is small enough to move the functionality inside this function.

---
 net/mac802154/rx.c | 12 +++---------
 1 file changed, 3 insertions(+), 9 deletions(-)
net/mac802154/rx.c +3 -9
index 04f3d61..246a60e 100644
@@ -242,11 +242,12 @@ mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 	rcu_read_unlock();
 }
 
-static void
-mac802154_subif_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
+void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
 {
 	struct ieee802154_local *local = hw_to_local(hw);
 
+	WARN_ON_ONCE(softirq_count() == 0);
+
 	skb->protocol = htons(ETH_P_IEEE802154);
 	skb_reset_mac_header(skb);
 
@@ -273,13 +274,6 @@ mac802154_subif_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
 fail:
 	kfree_skb(skb);
 }
-
-void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
-{
-	WARN_ON_ONCE(softirq_count() == 0);
-
-	mac802154_subif_rx(hw, skb);
-}
 EXPORT_SYMBOL(ieee802154_rx);
 
 void
Signature · 1 line
2.1.2
author maintainer
#8 lore raw

PATCH v2 8/16 mac802154: rx: move skb->protocol setting

In reply to Alexander Aring

This patch moves the skb->protocol setting to the position when it's needed. It's only needed when frame parsing was successful.

---
 net/mac802154/rx.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
net/mac802154/rx.c +4 -1
index 246a60e..c4066b5 100644
@@ -31,6 +31,8 @@
 
 static int mac802154_process_data(struct net_device *dev, struct sk_buff *skb)
 {
+	skb->protocol = htons(ETH_P_IEEE802154);
+
 	return netif_receive_skb(skb);
 }
 
@@ -224,6 +226,8 @@ mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 	u16 crc = crc_ccitt(0, skb->data, skb->len);
 	u8 *data;
 
+	skb->protocol = htons(ETH_P_IEEE802154);
+
 	rcu_read_lock();
 	list_for_each_entry_rcu(sdata, &local->interfaces, list) {
 		if (sdata->type != IEEE802154_DEV_MONITOR ||
@@ -248,7 +252,6 @@ void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
 
 	WARN_ON_ONCE(softirq_count() == 0);
 
-	skb->protocol = htons(ETH_P_IEEE802154);
 	skb_reset_mac_header(skb);
 
 	if (!(local->hw.flags & IEEE802154_HW_OMIT_CKSUM)) {
Signature · 1 line
2.1.2
author maintainer
#9 lore raw

PATCH v2 9/16 mac802154: rx: add CHECKSUM_UNNECESSARY

In reply to Alexander Aring

This patch adds CHECKSUM_UNNECESSARY to skb->ip_summed before delivery. There exist no transceiver with IP checksum functionality.

---
 net/mac802154/rx.c | 2 ++
 1 file changed, 2 insertions(+)
net/mac802154/rx.c +2 -0
index c4066b5..14eecac 100644
@@ -31,6 +31,7 @@
 
 static int mac802154_process_data(struct net_device *dev, struct sk_buff *skb)
 {
+	skb->ip_summed = CHECKSUM_UNNECESSARY;
 	skb->protocol = htons(ETH_P_IEEE802154);
 
 	return netif_receive_skb(skb);
@@ -226,6 +227,7 @@ mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 	u16 crc = crc_ccitt(0, skb->data, skb->len);
 	u8 *data;
 
+	skb->ip_summed = CHECKSUM_UNNECESSARY;
 	skb->protocol = htons(ETH_P_IEEE802154);
 
 	rcu_read_lock();
Signature · 1 line
2.1.2
author maintainer
#10 lore raw

PATCH v2 10/16 mac802154: rx: add monitor pkt_type information

In reply to Alexander Aring

This patch adds a PACKET_OTHERHOST setting when a monitor interface receives a skb. All receiving skb's to the monitor interface should be PACKET_OTHERHOST.

---
 net/mac802154/rx.c | 1 +
 1 file changed, 1 insertion(+)
net/mac802154/rx.c +1 -0
index 14eecac..f7f09b4 100644
@@ -228,6 +228,7 @@ mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 	u8 *data;
 
 	skb->ip_summed = CHECKSUM_UNNECESSARY;
+	skb->pkt_type = PACKET_OTHERHOST;
 	skb->protocol = htons(ETH_P_IEEE802154);
 
 	rcu_read_lock();
Signature · 1 line
2.1.2
author maintainer
#11 lore raw

PATCH v2 11/16 mac802154: rx: move skb_reset_mac_header

In reply to Alexander Aring

This patch moves the skb_reset_mac_header call before frame parsing while wpan rx and before monitor deliver functionality.

---
 net/mac802154/rx.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
net/mac802154/rx.c +3 -2
index f7f09b4..c299980 100644
@@ -138,6 +138,8 @@ static int mac802154_parse_frame_start(struct sk_buff *skb,
 	int hlen;
 	struct ieee802154_mac_cb *cb = mac_cb_init(skb);
 
+	skb_reset_mac_header(skb);
+
 	hlen = ieee802154_hdr_pull(skb, hdr);
 	if (hlen < 0)
 		return -EINVAL;
@@ -227,6 +229,7 @@ mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 	u16 crc = crc_ccitt(0, skb->data, skb->len);
 	u8 *data;
 
+	skb_reset_mac_header(skb);
 	skb->ip_summed = CHECKSUM_UNNECESSARY;
 	skb->pkt_type = PACKET_OTHERHOST;
 	skb->protocol = htons(ETH_P_IEEE802154);
@@ -255,8 +258,6 @@ void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
 
 	WARN_ON_ONCE(softirq_count() == 0);
 
-	skb_reset_mac_header(skb);
-
 	if (!(local->hw.flags & IEEE802154_HW_OMIT_CKSUM)) {
 		u16 crc;
 
Signature · 1 line
2.1.2
author maintainer
#12 lore raw

PATCH v2 12/16 mac802154: rx: move rcu locking

In reply to Alexander Aring

Instead of twice lock and unlock mechanism this patch hold these locks only once at one position.

---
 net/mac802154/rx.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
net/mac802154/rx.c +4 -4
index c299980..689bb7f 100644
@@ -205,7 +205,6 @@ mac802154_wpans_rx(struct ieee802154_local *local, struct sk_buff *skb)
 		return;
 	}
 
-	rcu_read_lock();
 	list_for_each_entry_rcu(sdata, &local->interfaces, list) {
 		if (sdata->type != IEEE802154_DEV_WPAN ||
 		    !netif_running(sdata->dev))
@@ -215,7 +214,6 @@ mac802154_wpans_rx(struct ieee802154_local *local, struct sk_buff *skb)
 		skb = NULL;
 		break;
 	}
-	rcu_read_unlock();
 
 	if (skb)
 		kfree_skb(skb);
@@ -234,7 +232,6 @@ mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 	skb->pkt_type = PACKET_OTHERHOST;
 	skb->protocol = htons(ETH_P_IEEE802154);
 
-	rcu_read_lock();
 	list_for_each_entry_rcu(sdata, &local->interfaces, list) {
 		if (sdata->type != IEEE802154_DEV_MONITOR ||
 		    !netif_running(sdata->dev))
@@ -249,7 +246,6 @@ mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 
 		netif_rx_ni(skb2);
 	}
-	rcu_read_unlock();
 }
 
 void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
@@ -273,9 +269,13 @@ void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
 		skb_trim(skb, skb->len - 2); /* CRC */
 	}
 
+	rcu_read_lock();
+
 	mac802154_monitors_rx(local, skb);
 	mac802154_wpans_rx(local, skb);
 
+	rcu_read_unlock();
+
 	return;
 
 fail:
Signature · 1 line
2.1.2
author maintainer
#13 lore raw

PATCH v2 13/16 ieee802154: add valid psdu length helper

In reply to Alexander Aring

This patch adds a generic valid psdu length check function helper. This is useful to check the length field after receiving. For example the at86rf231 doesn't filter invalid psdu length. Sometimes the CRC can also be correct. If we get the lqi value with an invalid frame length the kernel may crash because we dereference an invalid pointer in the receive buffer.

---
 include/linux/ieee802154.h | 11 +++++++++++
 1 file changed, 11 insertions(+)
include/linux/ieee802154.h +11 -0
index 2dfab2d..6e50a2a 100644
@@ -23,7 +23,10 @@
 #ifndef LINUX_IEEE802154_H
 #define LINUX_IEEE802154_H
 
+#include <linux/types.h>
+
 #define IEEE802154_MTU			127
+#define IEEE802154_MIN_PSDU_LEN		5
 
 #define IEEE802154_FC_TYPE_BEACON	0x0	/* Frame is beacon */
 #define	IEEE802154_FC_TYPE_DATA		0x1	/* Frame is data */
@@ -185,5 +188,13 @@ enum {
 	IEEE802154_SCAN_IN_PROGRESS = 0xfc,
 };
 
+/**
+ * ieee802154_is_valid_psdu_len - check if psdu len is valid
+ * @len: psdu len with (MHR + payload + MFR)
+ */
+static inline bool ieee802154_is_valid_psdu_len(const u8 len)
+{
+	return (len >= IEEE802154_MIN_PSDU_LEN && len <= IEEE802154_MTU);
+}
 
 #endif /* LINUX_IEEE802154_H */
Signature · 1 line
2.1.2
author maintainer
#14 lore raw

PATCH v2 14/16 at86rf230: use ieee802154_is_valid_psdu_len helper

In reply to Alexander Aring

This patch adds the ieee802154_is_valid_psdu_len function to validate the psdu length. If the psdu length is invalid we use the maximum payload to receive also corrupted frames in monitor mode.

---
 drivers/net/ieee802154/at86rf230.c | 26 ++++++++------------------
 1 file changed, 8 insertions(+), 18 deletions(-)
drivers/net/ieee802154/at86rf230.c +8 -18
index e4fbcaa..368791b 100644
@@ -788,26 +788,11 @@ at86rf230_tx_trac_status(void *context)
 
 static void
 at86rf230_rx(struct at86rf230_local *lp,
-	     const u8 *data, u8 len)
+	     const u8 *data, const u8 len)
 {
-	u8 lqi;
 	struct sk_buff *skb;
 	u8 rx_local_buf[AT86RF2XX_MAX_BUF];
 
-	if (len < 2)
-		return;
-
-	/* read full frame buffer and invalid lqi value to lowest
-	 * indicator if frame was is in a corrupted state.
-	 */
-	if (len > IEEE802154_MTU) {
-		lqi = 0;
-		len = IEEE802154_MTU;
-		dev_vdbg(&lp->spi->dev, "corrupted frame received\n");
-	} else {
-		lqi = data[len];
-	}
-
 	memcpy(rx_local_buf, data, len);
 	enable_irq(lp->spi->irq);
 
@@ -822,7 +807,7 @@ at86rf230_rx(struct at86rf230_local *lp,
 	/* We do not put CRC into the frame */
 	skb_trim(skb, len - 2);
 
-	ieee802154_rx_irqsafe(lp->hw, skb, lqi);
+	ieee802154_rx_irqsafe(lp->hw, skb, rx_local_buf[len]);
 }
 
 static void
@@ -831,7 +816,12 @@ at86rf230_rx_read_frame_complete(void *context)
 	struct at86rf230_state_change *ctx = context;
 	struct at86rf230_local *lp = ctx->lp;
 	const u8 *buf = lp->irq.buf;
-	const u8 len = buf[1];
+	u8 len = buf[1];
+
+	if (!ieee802154_is_valid_psdu_len(len)) {
+		dev_vdbg(&lp->spi->dev, "corrupted frame received\n");
+		len = IEEE802154_MTU;
+	}
 
 	at86rf230_rx(lp, buf + 2, len);
 }
Signature · 1 line
2.1.2
author maintainer
#15 lore raw

PATCH v2 15/16 at86rf230: improve receive handling

In reply to Alexander Aring

Current behaviour it to copy the frame inclusive CRC into a skb, then remove the CRC from the skb. This patch optimizes this by not copying the CRC at the first place.

---
 drivers/net/ieee802154/at86rf230.c | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)
drivers/net/ieee802154/at86rf230.c +3 -7
index 368791b..a155838 100644
@@ -788,7 +788,7 @@ at86rf230_tx_trac_status(void *context)
 
 static void
 at86rf230_rx(struct at86rf230_local *lp,
-	     const u8 *data, const u8 len)
+	     const u8 *data, const u8 len, const u8 lqi)
 {
 	struct sk_buff *skb;
 	u8 rx_local_buf[AT86RF2XX_MAX_BUF];
@@ -803,11 +803,7 @@ at86rf230_rx(struct at86rf230_local *lp,
 	}
 
 	memcpy(skb_put(skb, len), rx_local_buf, len);
-
-	/* We do not put CRC into the frame */
-	skb_trim(skb, len - 2);
-
-	ieee802154_rx_irqsafe(lp->hw, skb, rx_local_buf[len]);
+	ieee802154_rx_irqsafe(lp->hw, skb, lqi);
 }
 
 static void
@@ -823,7 +819,7 @@ at86rf230_rx_read_frame_complete(void *context)
 		len = IEEE802154_MTU;
 	}
 
-	at86rf230_rx(lp, buf + 2, len);
+	at86rf230_rx(lp, buf + 2, len - 2, buf[2 + len]);
 }
 
 static void
Signature · 1 line
2.1.2
author maintainer
#16 lore raw

PATCH v2 16/16 mac802154: rx: change naming convention

In reply to Alexander Aring

This patch changes the naming convention of mac802154 rx file. It should be more named like mac80211 stack. Furthermore we introduce a new frame parsing implementation which is much similar the mac80211 implementation.

---
 net/mac802154/rx.c | 33 +++++++++++++++++----------------
 1 file changed, 17 insertions(+), 16 deletions(-)
net/mac802154/rx.c +17 -16
index 689bb7f..86394be 100644
@@ -29,7 +29,7 @@
 
 #include "ieee802154_i.h"
 
-static int mac802154_process_data(struct net_device *dev, struct sk_buff *skb)
+static int ieee802154_deliver_skb(struct net_device *dev, struct sk_buff *skb)
 {
 	skb->ip_summed = CHECKSUM_UNNECESSARY;
 	skb->protocol = htons(ETH_P_IEEE802154);
@@ -38,8 +38,8 @@ static int mac802154_process_data(struct net_device *dev, struct sk_buff *skb)
 }
 
 static int
-mac802154_subif_frame(struct ieee802154_sub_if_data *sdata, struct sk_buff *skb,
-		      const struct ieee802154_hdr *hdr)
+ieee802154_subif_frame(struct ieee802154_sub_if_data *sdata,
+		       struct sk_buff *skb, const struct ieee802154_hdr *hdr)
 {
 	__le16 span, sshort;
 	int rc;
@@ -103,7 +103,7 @@ mac802154_subif_frame(struct ieee802154_sub_if_data *sdata, struct sk_buff *skb,
 
 	switch (mac_cb(skb)->type) {
 	case IEEE802154_FC_TYPE_DATA:
-		return mac802154_process_data(sdata->dev, skb);
+		return ieee802154_deliver_skb(sdata->dev, skb);
 	default:
 		pr_warn("ieee802154: bad frame received (type = %d)\n",
 			mac_cb(skb)->type);
@@ -115,8 +115,8 @@ fail:
 	return NET_RX_DROP;
 }
 
-static void mac802154_print_addr(const char *name,
-				 const struct ieee802154_addr *addr)
+static void
+ieee802154_print_addr(const char *name, const struct ieee802154_addr *addr)
 {
 	if (addr->mode == IEEE802154_ADDR_NONE)
 		pr_debug("%s not present\n", name);
@@ -132,8 +132,8 @@ static void mac802154_print_addr(const char *name,
 	}
 }
 
-static int mac802154_parse_frame_start(struct sk_buff *skb,
-				       struct ieee802154_hdr *hdr)
+static int
+ieee802154_parse_frame_start(struct sk_buff *skb, struct ieee802154_hdr *hdr)
 {
 	int hlen;
 	struct ieee802154_mac_cb *cb = mac_cb_init(skb);
@@ -153,8 +153,8 @@ static int mac802154_parse_frame_start(struct sk_buff *skb,
 	cb->ackreq = hdr->fc.ack_request;
 	cb->secen = hdr->fc.security_enabled;
 
-	mac802154_print_addr("destination", &hdr->dest);
-	mac802154_print_addr("source", &hdr->source);
+	ieee802154_print_addr("destination", &hdr->dest);
+	ieee802154_print_addr("source", &hdr->source);
 
 	cb->source = hdr->source;
 	cb->dest = hdr->dest;
@@ -192,13 +192,14 @@ static int mac802154_parse_frame_start(struct sk_buff *skb,
 }
 
 static void
-mac802154_wpans_rx(struct ieee802154_local *local, struct sk_buff *skb)
+__ieee802154_rx_handle_packet(struct ieee802154_local *local,
+			      struct sk_buff *skb)
 {
 	int ret;
 	struct ieee802154_sub_if_data *sdata;
 	struct ieee802154_hdr hdr;
 
-	ret = mac802154_parse_frame_start(skb, &hdr);
+	ret = ieee802154_parse_frame_start(skb, &hdr);
 	if (ret) {
 		pr_debug("got invalid frame\n");
 		kfree_skb(skb);
@@ -210,7 +211,7 @@ mac802154_wpans_rx(struct ieee802154_local *local, struct sk_buff *skb)
 		    !netif_running(sdata->dev))
 			continue;
 
-		mac802154_subif_frame(sdata, skb, &hdr);
+		ieee802154_subif_frame(sdata, skb, &hdr);
 		skb = NULL;
 		break;
 	}
@@ -220,7 +221,7 @@ mac802154_wpans_rx(struct ieee802154_local *local, struct sk_buff *skb)
 }
 
 static void
-mac802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
+ieee802154_monitors_rx(struct ieee802154_local *local, struct sk_buff *skb)
 {
 	struct sk_buff *skb2;
 	struct ieee802154_sub_if_data *sdata;
@@ -271,8 +272,8 @@ void ieee802154_rx(struct ieee802154_hw *hw, struct sk_buff *skb)
 
 	rcu_read_lock();
 
-	mac802154_monitors_rx(local, skb);
-	mac802154_wpans_rx(local, skb);
+	ieee802154_monitors_rx(local, skb);
+	__ieee802154_rx_handle_packet(local, skb);
 
 	rcu_read_unlock();
 
Signature · 1 line
2.1.2
#17 lore raw
In reply to Alexander Aring

Hi Alex,

Quoted 38 lines

This patch serie improves and cleanups the receive handling. Instead workqueue we use a tasklet now for parsing frames. The first patch is a fix for using put_unaligned_le16 to copy checksum in skb data which was reported by Marcel Holtmann.

Other patches are cleanups for preparing further patches to introduce a frame parsing like mac80211. We don't change the printout function to netdev_foo now because the most code will be replaced by the new frame parsing implementation.

changes since v2:
- correct ieee802154 instead ieee80211 in patch 04/16 ("mac802154: rx:
document ieee802154_rx() context requirement")
- reword complete commit message of patch 04/16 ("mac802154: rx: document ieee802154_rx() context requirement")
- fix && instead || in patch 13/16 ("ieee802154: add valid psdu length helper")
- make patch 13/16 ("ieee802154: add valid psdu length helper") more human readable to check start at first end then end interval. Instead end <-> start checking.

Alexander Aring (16):
mac802154: tx: use put_unaligned_le16 for copy crc
ieee802154: drivers: use dev_alloc_skb
mac802154: rx: use tasklet instead workqueue
mac802154: rx: document ieee802154_rx() context requirement
mac802154: rx: move receive handling into rx.c
mac802154: tx: remove monitor receive while xmit
mac802154: rx: rename remove mac802154_subif_rx
mac802154: rx: move skb->protocol setting
mac802154: rx: add CHECKSUM_UNNECESSARY
mac802154: rx: add monitor pkt_type information
mac802154: rx: move skb_reset_mac_header
mac802154: rx: move rcu locking
ieee802154: add valid psdu length helper
at86rf230: use ieee802154_is_valid_psdu_len helper
at86rf230: improve receive handling
mac802154: rx: change naming convention

Quoted 12 lines
drivers/net/ieee802154/at86rf230.c |  32 ++---
drivers/net/ieee802154/cc2520.c    |   2 +-
drivers/net/ieee802154/mrf24j40.c  |   2 +-
include/linux/ieee802154.h         |  11 ++
include/net/mac802154.h            |   1 +
net/mac802154/ieee802154_i.h       |   9 +-
net/mac802154/iface.c              | 186 -------------------------
net/mac802154/main.c               |  30 ++++
net/mac802154/monitor.c            |  27 ----
net/mac802154/rx.c                 | 279 +++++++++++++++++++++++++++++++------
net/mac802154/tx.c                 |   7 +-
11 files changed, 296 insertions(+), 290 deletions(-)

all 16 patches have been applied to bluetooth-next tree.

Regards

Marcel